In Firefox with CHTML rendering, issues with speech and accessibility explorer

5 views
Skip to first unread message

Alex Jordan

unread,
Jan 31, 2026, 5:03:15 PMJan 31
to MathJax Users
With Chrome, I go to this page:
which is a PreTeXt sample page. It is using MathJax 3.

I access the MJ menu, confirm CHTML rendering, Activate Accessibility, and reload the page. With focus on the first piece of display math (the quadratic formula), I use shift+space, and I have entered the explorer. This is all good and as I have expected.

Now I repeat this again with Firefox 147.0.2 using Mac OS Tahoe 26.2. (Worth noting that the OS is a recent upgrade and it seems to me that some issues started happening for me with this upgrade.) When I get to focus on the display math, shift+space does not enter the explorer. It exhibits the default browser behavior for shift+space, which is to scroll up some.

A strange thing is that in the MJ menu, none of the speech rules are selected. If I select any, the selection does not stick:

Screenshot 2026-01-31 at 1.55.57 PM.png


This is true with the inline math expressions as well (no speech rules selected), although with the inline math, I do not have the issue with the shift+space behavior.

If I switch to SVG rendering, I do not have the shift+space issue nor the speech rule selection issue.

FWIW, I feel like this behavior is new, and possibly related to my OS upgrade or a recent Firefox upgrade. This all seemed a bit much to report on GitHub, so I thought I'd start here to see if this is known. And if version 3 is even maintained at this point.


Davide Cervone

unread,
Feb 2, 2026, 6:35:07 AMFeb 2
to mathja...@googlegroups.com
Alex:

Y9ou are correct that the speech rule choice is not being properly handled, and it doesn't show in the menu properly and the saved value isn't being used to set the speech rules that are used.  The reason is that the speech rules actually use two separate preferences that are combined into one saved preference, and the v3 code doesn't combine/decompose that properly.

Here is a configuration that patches the menu to properly handle the speech rule settings:

MathJax = {
  loader: {
    'ui/menu': {
      ready() {
        const {Menu} = MathJax._.ui.menu.Menu;
        Object.assign(Menu.prototype, {
          _applySettings_: Menu.prototype.applySettings,
          applySettings() {
            const settings = this.settings;
            const sre = this.document.options.sre;
            if (!settings.speechRules) {
              settings.speechRules = `${sre.domain || 'mathspeak'}-${sre.style || 'default'}`;
            }
            const [domain, style] = settings.speechRules.split('-');
            sre.domain = domain;
            sre.style = style;
            this._applySettings_();
          },
          _getA11y_: Menu.prototype.getA11y,
          getA11y(name) {
            return name === 'speechRules'
               ? this.settings.speechRules
               : this._getA11y_(name);
          },
          _setA11y_: Menu.prototype.setA11y,
          setA11y(options) {
            options = {...options};
            delete options.speechRules;
            this._setA11y_(options);
          }
        });
      }
    },
  },
};

I haven't updates to Tahow yet, so I can' test Firefox there, but it works for me in Sequoia 15.5.  So that does seem to be new behavior.  Pressing ENTER should also enter the explorer.  Does that work for you in Tahoe?

Davide


On Jan 31, 2026, at 5:02 PM, Alex Jordan <jordanc...@gmail.com> wrote:

With Chrome, I go to this page:
which is a PreTeXt sample page. It is using MathJax 3.

I access the MJ menu, confirm CHTML rendering, Activate Accessibility, and reload the page. With focus on the first piece of display math (the quadratic formula), I use shift+space, and I have entered the explorer. This is all good and as I have expected.

Now I repeat this again with Firefox 147.0.2 using Mac OS Tahoe 26.2. (Worth noting that the OS is a recent upgrade and it seems to me that some issues started happening for me with this upgrade.) When I get to focus on the display math, shift+space does not enter the explorer. It exhibits the default browser behavior for shift+space, which is to scroll up some.

A strange thing is that in the MJ menu, none of the speech rules are selected. If I select any, the selection does not stick:

<Screenshot 2026-01-31 at 1.55.57 PM.png>


This is true with the inline math expressions as well (no speech rules selected), although with the inline math, I do not have the issue with the shift+space behavior.

If I switch to SVG rendering, I do not have the shift+space issue nor the speech rule selection issue.

FWIW, I feel like this behavior is new, and possibly related to my OS upgrade or a recent Firefox upgrade. This all seemed a bit much to report on GitHub, so I thought I'd start here to see if this is known. And if version 3 is even maintained at this point.



--
You received this message because you are subscribed to the Google Groups "MathJax Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mathjax-user...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/mathjax-users/CA%2BR-jrfGbCGxTZ1e9nzLvh-%3DSozwvsf7_aoCqGFkkbtsrPsTKw%40mail.gmail.com.

Alex Jordan

unread,
Feb 2, 2026, 4:03:06 PMFeb 2
to mathja...@googlegroups.com
Thanks Davide!

My school manages my laptop OS and remotely upgraded me to Tahoe spontaneously one day without warning (rude). I can't 100% be sure that this upgrade is at fault, but it seems like since that upgrade, when using Firefox
(a) this shift+space behavior with MathJax started.
(b) in the WeBWorK problem editor, when I have focus in CodeMirror, I can no longer use control+return to render the problem.

> Pressing ENTER should also enter the explorer.  Does that work for you in Tahoe?

No. Same conditions: HTML rendering and with focus on display math. Using shift+enter scrolls up a bit, and enter does nothing. (Enter will enter the explorer if it's inline math or SVG rendering, or with Chrome.)


Reply all
Reply to author
Forward
0 new messages